Correctly calculating impedance of a biconical antenna and impedance matching

[dazz1]

Pretty poor though (10dB?).

The poor common mode attenuation was expected because of the low number of turns.  I was testing the hypothesis that reducing the cable length would increase the bandwidth.  The results prove that to be true.

Got any clip-on ferrites? 

Lots of them, the split clip-on type.   I was planning on using them on the coax that connects the antenna to the receiver.  The aim being to attenuated re-radiation of external signals picked up by the coax shield. 

This is probably the range where you need a single-turn winding on a loooong core.  Alternately, make a new cable with beads all up on it before crimping the connectors.

Tim

I haven't finished looking at all of the options for a coax/toroid balun yet.    I plan to try RG178 coax.  At 2mm diameter, it will be easier to wind tightly through the smaller toroid. That will reduce stray (differential) inductance to minimise insertion loss.  I will be able to wind more turns.  There will be less inter-winding capacitance.  The length of the coax will be less for a given number of turns.   The coax losses will be greater, but the insertion loss should still be low for such a short length.

The fixture for common mode testing has proven to work well except for the short cables that connect to the SA TG out and RF in.  I am thinking of making a new fixture that connects directly to the SA connectors.   

I don't make up my own coax assemblies.  It is far cheaper, easier and slower to buy them ready made from China.    That's what I plan to do for the RG178 assemblies. They won't arrive until some time around December so while I am waiting, I will work on another project.

[dazz1]

I have reason to suspect that my simple common mode test fixture is distorting the measurements because of the lengths of coax resonating.
The fixture links the inner signal conductors to/from the spectrum analyser to the outer conductors of the DUT.  It provides a direct measure of the attenuation of common mode currents on the outer shield of the DUT.  I have not seen this test fixture described in any literature.  There is probably a good reason.  Either it is a dumb idea that doesn't work very well, or I am a genius.  Unlikely to be the latter.

Given the apparent success of the simple CM test fixture I have made an improved version.  This version connects directly to the front panel connectors of my spectrum analyser (SA).  Very short coax then connects the input/output connectors.

The fixture is based on aluminium clad composite.  A black plastic core is sandwiched between two layers of aluminium.

The fixture includes a calibration assembly.  The calibration assembly makes the shortest possible conductor between the two SMA DUT connectors.  The SA normalises against this calibration assembly.  The assembly only connects the connector shields.  There is no signal on the inner coax conductor. 

I need a second N-type to SMA adapter before I can fully test the new CM test fixture.

I have considered using a twisted pair to wind the toroid but I cannot find short lengths of cable that would produce a 50ohm impedance.   I could use two Ethernet pairs (100ohm each) in parallel.    The coax measurements show low insertion loss, good matching and easy to build. 

[T3sl4co1l]

Note that you're making a 1/4 wave monopole here, very roughly speaking: you have unbalanced metal above a ground plane, coupled to a signal line.  Radiation depends on impedance along the length of the thing, so it's not going to be a good antenna with a lot of turns and ferrite in there, but the fact remains it's a blob of mostly-conductive stuff hanging out, and there will be radiation resistance as part of the measurement.

What's more, because it's unbalanced, and the "plane" is small, there will be plenty of feedline currents -- you should see a few dB of dependence on their position and length.

For best results, extending those "planes" to a full metal enclosure around the DUT would be a good idea, to contain those CM (displacement) currents, eliminate radiation (it's reflected back to the DUT), and improve consistency.  (There should be a peak/notch evident, corresponding to reflection from the enclosure -- cavity modes; these can be ignored, or if the DUT doesn't need to be too big, a small enclosure can be used to push it up higher.)

Tim

[dazz1]

Note that you're making a 1/4 wave monopole here, very roughly speaking: you have unbalanced metal above a ground plane, coupled to a signal line.  Radiation depends on impedance along the length of the thing, so it's not going to be a good antenna with a lot of turns and ferrite in there, but the fact remains it's a blob of mostly-conductive stuff hanging out, and there will be radiation resistance as part of the measurement.

I saw coax/antenna type resonance in the 1st version of the CM test fixture. The calibration link was a 15cm sma coax assembly plus the coax connections to the SA that acted like an antenna.   I could see external radio signals in the plots.   The quarter wave length of the entire test fixture was well inside the DUT frequency band of interest.  I reached a point in the testing where I could not differenciate between DUT characteristics and measurement errors. 

The ver 2 fixture is as short as practical .  The calibration link length is equivalent to a 1/4 wavelength antenna @ about 1.5GHz, well outside the 0Hz-300MHz frequency band of interest.   

What's more, because it's unbalanced, and the "plane" is small, there will be plenty of feedline currents -- you should see a few dB of dependence on their position and length.

For best results, extending those "planes" to a full metal enclosure around the DUT would be a good idea, to contain those CM (displacement) currents, eliminate radiation (it's reflected back to the DUT), and improve consistency.  (There should be a peak/notch evident, corresponding to reflection from the enclosure -- cavity modes; these can be ignored, or if the DUT doesn't need to be too big, a small enclosure can be used to push it up higher.)

Tim

I agree that enclosing the fixture would eliminate pickup of external signals (eg. cell phones, radio stations etc) but would add cavity reflections.  I have access to a local shielded chamber (not anechoic) that I would use if necessary.    Building a shielded fixture would be the option of last resort. 

My expectation is that the version 2 fixture should be "good enough" to make measurements that have sufficient certainty to make informed design decisions.   I will need to do some testing to find its limits.   Given my modest range of test equipment, I think this is the best I can do with what I have. 

[A.Z.]

Pretty poor though (10dB?). Got any clip-on ferrites?  This is probably the range where you need a single-turn winding on a loooong core.  Alternately, make a new cable with beads all up on it before crimping the connectors.

Tim

Agreed, a decent W2DU style choke/balun should work pretty well at those frequencies, just a matter of choosing the right material(s)  and adjusting the number of elements

[dazz1]

None of the references found for the W2DU balun, or any other type, provide a method for measuring the (im)balance of a balun.

It could be done by applying a signal generator to the unbalanced side, then connecting the balanced outputs to the X-Y inputs of an oscilloscope or VNA.  A perfect balun would display a 45 degree line/vector.
My scopes are rated to 100MHz and not the 300MHz I want to measure.

[tautech]

My scopes are rated to 100MHz and not the 300MHz I want to measure.

I can help with that. 
Down your ways in a couple weeks but without wheels if you can collect from Karori.

[dazz1]

I can help with that. 
Down your ways in a couple weeks but without wheels if you can collect from Karori.

It would be good to catch up again.    You can tell me all about the latest Siglent test equipment I want, but can't justify or afford. 
Right now, I can't get my balun to work to 300MHz, so even if I had the scope, it wouldn't help.
It is the usual conflict between wants and needs.   

I am no balun expert, but as I have learned more about their actual measured performance, it leads me to have doubts about claims of baluns I see on the Internet.  No measured data, flawed measurement methodology, and other issues raise these doubts.  To avoid the same issues I have invented or modified measurement methods to fit within the capabilities of my modest inventory of test equipment. Necessity being the mother of invention and all that.   So far the data is telling me that the critical part of an EMC antenna is the balun, and not the antenna.   

It appears that the most important characteristic of the balun is balance.   Without balance, the beam pattern will be distorted and the measured values of emission will be entirely unreliable.   Not good for an EMC antenna.    Measuring balance is made to be difficult by the fact that most (nearly all?) RF test equipment is unbalanced. 

I have ordered a N-type to SMA adapter from China, but that is not scheduled to arrive until Jan 2024.  I won't be able to fully test my new RF imbalance fixture until then.    Progress on this project is in burst mode.

[dazz1]

OK I have made another small step forward.  The new direct screw-on common mode test fixture is now complete.

As a reminder, this test fixture takes the unbalanced signal from the connector centre pin of the tracking generator output, and puts that signal onto the shield of the coax of the Device Under Test (DUT).  The signal is then shifted back onto the centre pin of the input of the spectrum analyzer.
The DUT is a toroid with coax windings to make, in this case, a 1:1 balun. 

Balanced currents within the coax are not affected by the toroid core.    Balanced currents have a low insertion loss through the coax and toroid windings.

Any imbalance in the balun results in common mode currents flowing on the outside of the coax shield.    These are affected by the ferrite toroid.  The toroid presents a high impedance to the exterior common mode currents, especially compared to the parallel path along the inside of the coax, balanced with the currents flowing on the coax centre wire. 

The test fixture applies an unbalanced current to the exterior of the DUT coax/ferrite core to measure the impedance and insertion loss.    High loss (>20dB) is good because that blocks common mode current (imbalance) and allows differential currents (balanced) to flow.  No signal is applied to the centre conductor of the DUT coax. 

The aim of building the Mk2 version of the CM test fixture was to minimise parasitics, including pickup of local transmitters in the area.  The fixture includes a reference link that is as close as practical to being a short circuit. 

Given the modest range of test equipment I have, this is the best method of measuring common mode current I can devise.  I have not seen this method applied by anyone anywhere else but I would be surprised if someone else has not done something the same way.

The attached plot shows the insertion loss of the CM test fixture compared to a short length of coax.    The coax was connected and the spectrum analyser was normalised.  The test fixture was then fitted, complete with shorting link.  The attached image shows the insertion loss out to 500MHz but performance remains good to much higher freqs.  The loss is low, and notably, there is no sign of any external rf Tx pickup.  Nor is there any sign of resonance or other bad behaviour.

The plot shows the new improved test fixture achieved the design aims.  No resonance.  Very low insertion loss.  No detectable external interference.

[dazz1]

With the new test fixture and a thin coax assembly wound on the smaller toroid, the attached plot was produced.
The attached plot shows acceptable performance out to 200MHz.    There is no indication of any resonance or other bad behaviour. 

This approach looks promising.  I have made no attempt to optimise this DUT.    The next step will be to reduce the number of windings to trade attenuation for bandwidth.    There are other things I can do, like change the ferrite material, or add very small toroid beads over the coax. 

[dazz1]

I thought I would investigate a coil with no ferrite.
I wound a few turns around a piece of PTFE just see what would happen.
As you might expect, the coil performance was poor.  No resonance, but not much attenuation either.

[dazz1]

And I also tried reducing the turns on the ferrite toroid.

[dazz1]

Next test was to retest my "reference" balun with the new text fixture.   My reference balun is just an arbitrary design that I use to compare with any other design.

[dazz1]

This test reduces the number of turns with the intention of reducing parasitics.  The aim being to increase the 20dB bandwidth out to 300MHz.
The trade off is that the  lowest frequency increases.

I may be able to improve the lower frequency attenuation by changing the toroid material.  This is still the best prototype yet.  It maintains >20dB attenuation up to about 350MHz.  Performance above 350MHz is well behaved.

I think I am now at the knee in the curve leading to diminishing returns.  Any significant changes might lead to small improvements.  I would like to improve the low frequency bandwidth.    Looking at the datasheets, changing to the Fair-rite grade 53 might achieve an improvement without wrecking the upper bandwidth.

[dazz1]

I thought I better just confirm the insertion loss with the "new" thin coax I am now using.
As you can see from the plot, the insertion loss is flat and insignificant.  This proves that the differential balanced current flowing inside the coax is not affected by the windings through the toroid.

I do have a balanced bridge, but it would not provide a direct reading of insertion loss.  The balanced bridge introduces measurement errors.

[dazz1]

I have thought about my next move.
I think I have gone as far as I can with what I have on-hand, which just happened to be about the right size ferrite toroid and the right material (61).

When I look at the Fair-rite website, ( https://fair-rite.com/materials/ ) the material 31 is recommended for a frequency range of 1-300MHz, exactly the range I want. 
The other material properties vary widely compared to 61.  I need to figure out what effects these material differences will have on balun core size and turns etc. 

Then I need to order the parts.  I need to order enough parts to avoid paying shipping costs.  That means there will be another pause in progress on this project until I need to order enough parts.

[tautech]

I have in the past requested a free sample.......
https://fair-rite.com/request-a-sample/

[dazz1]

I have in the past requested a free sample.......
https://fair-rite.com/request-a-sample/

Free is always good.  I will try that.

I have been considering ways of further testing the balun.  I know the text book method is with a VNA but I don't have one of those and, in this particular case, I don't think it is the best method.
The attached drawing shows the proposed test setup.  This is to illustrate the concept.  The final circuit will need more stuff.      This test setup will provide a direct visual comparison of imbalance in phase and amplitude in a way that a normal Smith chart can't.  I know that use of Lissajous patterns is far older than I, but the availability of wide bandwidth oscilloscopes makes them usable at frequencies that previously required specialist RF equipment, like a VNA.

The reason why I don't think a VNA is the right instrument for this type of test is that typical VNAs are two port devices.  The proposed setup is a 3 port system, with direct read out of imbalance.  In addition to phase and amplitude imbalance, it will also show losses (insertion loss).  A perfect lossless balun will display a straight 45 degree line.  Any loss will display hysteresis with an internal area proportional to losses. 

If I sweep the frequency, I will be able to produce a Smith chart showing the difference in phase and amplitude between the two ports.  The main issue with this setup is that the comparison is linear, not logarithmic.  That will adversely affect the dynamic range of the displayed comparison.

I don't have a 300MHz (~500MHz) scope, but I may be able to get access to one a short distance from where I live. 
My ancient Wavetek 2520A RF signal generator has GPIB so I can sweep the frequency.  It has been calibrated to 0.1dB across the whole frequency range.

If you think there is a better way of doing this, let me know.

[tautech]

You could always measure one leg of the balun with your analyzer and save it as a reference trace to compare the other balan leg against. 

Or failing that, View (freeze) the analyzer result and activate a 2nd trace (C&W) for the other leg of the balan.

Or when I'm down next time with wheels bring along a 4 port VNA.....

Looking forward to that coffee you suggested we have in a couple of weeks. 

[dazz1]

Or when I'm down next time with wheels bring along a 4 port VNA.....

If you did that, I would just drool all over it like my dog waiting for dinner . 

[dazz1]

Based on this video:
https://www.youtube.com/watch?app=desktop&v=SuOEWapx62c
it would seem that my ancient 100MHz HP54645D is usable out to about 300MHz, the planned bandwidth of the biconic antenna.  The amplitude is attenuated, and no doubt there will be phase shift, but for my purposes, this should be indicative.    So long as both DSO channels have the same amplitude and phase shift, I should still see usable Lissajous patterns.

[dazz1]

Fair-Rite have come to the party and offered samples so my R&D can continue.   

I have asked for a toroid of similar dimensions to the part I already have (5961001701), type 61 material.
The key difference with the requested part (2631801202) is the material type 31. 

The µ for 31 is 10x greater than for 61. 10 turns on a 61 toroid should be the same as 1 turn on a 31 toroid.
or
10 turns on a 31 toroid should provide a lower frequency -20dB point than a 61 toroid. 

The penalty is that 31 has a lower upper frequency limit, but that should be OK.  The type 61 toroid balun I have developed already has a greater bandwidth than the antenna I have built.  The balun has a wider bandwidth than equivalent commercial biconic antennae.  So you might wonder why I don't stop and have a beer in celebration of my self-acknowledged genius 

So while the theory tells me the 31 material will be a Cinderella solution, the practice might be a little different.  I anticipate I will need to play around with the number of turns to trade bandwidth for attenuation.  Just like before. 

I have built two identical antenna so I can calibrate them  Once I have done that, one antenna will be spare.   I was planning on selling the second biconic but I doubt anyone will be willing to pay enough for it.   If anyone was to ask the price for me to make/sell  one of these biconics, that question alone would suggest they can't afford it.  So far I have a waiting list of zero.   

If I keep it,  I can very easily and cheaply make another set of elements that plug-in to create a biconic with a different frequency range.  That will be a lot easier if I can make an ultra-wideband balun so I can change-out elements without changing the balun.   

The cost/effort of trying the 31 toroid is cheaper/easier/faster than making another set of hubs. 

I have started to order the parts for my cheap-as 3 port, X-Y balun balance tester.  It might take a while for those items to get through the Christmas crush of presents clogging the mail system.

The more I learn about baluns, the more skeptical I am of the performance of some of the commercial biconics I see advertised. Any unbalance in the performance of the balun will distort the beam pattern.  Few, if any, antenna datasheets include a plot of the actual measured 3D beam pattern.  I have seen simulated beam patterns, no doubt based on theoretically perfect baluns.    Some might say it doesn't matter because the antenna only has to receive in one direction.  That chain of thought ignores the effects of the beam not pointing where it should.  If I was in the market for an EMC antenna, I would be focusing on the measured beam pattern and not relying on the data sheet and brand name. 



I am expecting another pause on this project until I get the 31 toroids and build the 3-port balun balance tester.   

Fluffy cats and cute pets always attract lots of views so here are mine.

[dazz1]

I thought I would measure the workable bandwidth of my Phillips PM3070 100MHz scope.  I also wanted to test if I needed amplifiers in my 3 port balun balance tester.

I connected up my ancient Wavetek 2520 RF sig gen to the Phillips scope.  The connection to the scope was terminated with a 50ohm load adapter to avoid false readings from reflections. 

The photo shows the scope display with a 13dBm input at 250MHz.    This is the end limit.  At any higher frequency, it would not trigger.

I found that the trigger was the weakest link.  As the frequency increased, the signal amplitude had to be increased to trigger.
At 250MHz there was significant amplitude reduction.  For my purposes, this is of no consequence because I will be looking at the X-Y phase relationship in Lissajous plots.   
There will undoubtedly be phase shift through the scope circuits, but as long as the  lag is the same on both channels, no problem.

So the Mod 1 version of my balun balance tester should work comfortably to 200MHz with the Phillips scope. 

I ran similar tests with the HP 54645D 100MHz scope.  As expected, there was significant amplitude roll-off at higher-than-spec frequencies.
The HP 54645D could comfortably display 250MHz signals  and still workable at 300MHz (but a little noisy). 
This scope reached the limits 330MHz @ 13dBm.  The displayed signal was noisy and the scope could burst into full on aliasing.

For both vintage scopes the test results are impressive and more than adequate for my balun balance tester (BBT).  I will not need amplifiers in the circuit.  It seems that, as usual with the RF things I make, the performance of the BBT will be defined by the physical construction.  No electronics required. 

[dazz1]

I received the parts I needed for my 3 port balun balance tester.
These included a pair of identical 50R BNC > BNC terminations to plug into the scope inputs.
For test purposes, a Tee was the DUT.    This has perfect balance.

This test was to figure out the limits of the tester to identify measurement errors and just see if it works. 

The first attachment shows the xy plot at 100MHz.  The 45 degree diagonal trace shows perfect balance.  Any loss would appear as a loop rather than a straight line.

At 350MHz, 13dBm the scope struggles but gives a stable, readable and useful trace.  Notice here that the line is no longer at 45 degrees indicative of an imbalance.   Some experimentation showed that one channel has slightly different gain to the other.  As long as I know this, I can compensate for it. 

In the xy mode, there is no trigger so I could still see a small but just usable trace a 400MHz@13dBm.  Not bad for a 100Mhz scope.

The next step is to configure the balun for 3 ports.    These are:
1.   signal and shield to RF sig-gen output
2.   signal to scope X input
3.   shield to scope Y input.

If the balun is doing its job, the X-Y trace should remain at 45 degrees with no loop across the frequency range.

The Fair-rite offer of free parts is too expensive to accept. Free parts, yes.  Free shipping, no.
It is cheaper for me to pay for the parts and get free shipping from Digi-key, element14 etc.

[dazz1]

There is an interesting post here:  https://www.eevblog.com/forum/rf-microwave/diy-rf-emc-biconic-antenna/75/ showing the balun for a TekBox Biconic antenna.
What would be more interesting is to measure the performance of this balun.